The impact of Fama French factors and coskewness in ISE

(1)

DOKUZ EYLÜL UNIVERSITY

GRADUATE SCHOOL OF SOCIAL SCIENCES DEPARTMENT OF BUSINESS ADMINISTRATION

FINANCE MASTER’S PROGRAM MASTER’S THESIS

THE IMPACT OF FAMA-FRENCH FACTORS AND

COSKEWNESS IN ISE

Fatma DEMRCAN

Supervisor

Prof. Dr. Aye Tülay YÜCEL

(2)

(3)

ii

DECLARATION

I hereby declare that this master’s thesis titled as “The Impact of Fama-French Factors and Coskewness in ISE” has been written by myself without applying the help that can be contrary to academic rules and ethical conduct. I also declare that all materials benefited in this thesis consist of the mentioned resourses in the reference list. I verify all these with my honour.

Date

Fatma DEMİRCAN Signature

(4)

iii

ÖZET Yüksek Lisans Tezi

Fama French Faktörlerinin ve Piyasa Faktörünün İmkb’deki Etkisi

Fatma Demircan

Dokuz Eylül Üniversitesi Sosyal Bilimler Enstitüsü İngilizce İşletme Anabilim Dalı

İngilizce Finansman Programı

Varlık fiyatlandırma yatırımcılar ve finans bilim adamları için her zaman ilgi çekici bir konu olmuştur. Varlık fiyatlandırma kavramında, Sermaye Varlıklarını Fiyatlandırma Modeli (SVFM) önemli bir yere sahiptir. SVFM, geliştirildiği günden itibaren finans dünyasında büyük ilgi görmüştür. Model hala portföy yönetimi ve sermaye maliyetinin tahminlemesi vb. akademik çalışma ve uygulamalarda yaygın olarak kullanılmaktadır. SVFM, gerçekçi olmayan varsayımlarından dolayı çok sayıda eleştiri almıştır. Modelin geliştirilmesinden günümüze kadar, modelin çeşitli türevleri ortaya çıkmıştır.

Bu tez çalışmasında, Sermaye Varlıklarını Fiyatlandırma Modeli, Fama French Üç Faktör Modeli ve Fama French Üç Faktör Modeli’ ne piyasaya göre çarpıklık faktörünün eklenmesiyle oluşan Dört Faktör Modeli uygulanmıştır. Temmuz 2002- Haziran 2010 döneminde, piyasaya göre çarpıklık faktörünün endüstri, büyüklük, defter değeri/piyasa değeri ve piyasa değeri ve momentuma göre İMKB’ de yer alan hisselerden oluşturulmuş olan portföylerin değişkenliğindeki etkisi, Harvey ve Siddique (2000)’ nun metodolojisine benzer bir yöntem kullanılarak incelenmiştir. Daha sonra, piyasaya göre çarpıklık faktörü ve Fama French faktörlerinin piyasa faktörünün portföylerin artık getirilerinin değişkenliğini açıklama gücünü arttırıcı bir etkisinin olup olmadığı Gibbons–Ross–Shanken (1989)’ in çok değişkenli testi ile test edilmiştir. Son

(5)

iv

olarak, incelenen varlık fiyatlandırma modellerinde yer alan faktörlerin açıklama gücü Fama- MacBeth ve Full Information Maximum Likelihood (FIML) metotları ile test edilmiştir.

Anahtar Kelimeler: Varlık Fiyatlandırma, Piyasaya Göre Çarpıklık, Fama French

(6)

v

ABSTRACT Master’s Thesis

The Impact of Fama French Factors and Coskewness in ISE Fatma Demircan

Dokuz Eylül University Graduate School of Social Sciences Department of Business Administration

Finance Master’s Program

Asset pricing is always an attractive topic for investors and finance scholars. The Capital Asset Pricing Model (CAPM) has an important place in the concept of asset pricing and it has attracted a great deal of attention by the finance world since it was developed. It is still widely used in academic research and applications, such as portfolio management and estimation of the cost of capital. The CAPM has taken a large number of criticism because of its unrealistic assumptions and from the time of the creation of the model to the present, different versions of the model have emerged

In this dissertation, the CAPM, the Fama French Three Factor model and the Four Factor Model which incorporates skewness – third order co-moment- into the Fama French Three Factor model are implemented. The impact of coskewness on the variation of portfolios that are formed according to industry, size, book to market ratio and momentum for ISE over the period July 2002 to June 2010 are investigated by using a similar methodology to Harvey and Siddique (2000). Then by performing multivariate testing procedure of Gibbons–Ross–Shanken (1989), these models are tested to reveal whether there is an incremetal effect of coskewness and Fama French factors to the market factor in explaining the variations of excess returns. Lastly, the explanatory power of factors in the analysed asset pricing models are tested by means of the Fama- MacBeth and Full Information Maximum Likelihood (FIML) methods.

(7)

vi

(8)

vii

THE IMPACT OF FAMA-FRENCH FACTORS AND COSKEWNESS IN ISE

INDEX

THESIS APPROVAL SHEET

DECLARATION ii ÖZET iii ABSTRACT v INDEX vii ABBREVIATIONS ix LIST OF TABLES xi LIST OF FIGURES xii

INTRODUCTION 1

CHAPTER ONE ASSET PRICING MODELS 1.1 THE CAPITAL ASSET PRICING MODEL 3

1.2 MULTI-FACTOR MODELS 6

1.2.1 The Arbitrage Pricing Theory 7

1.2.2 Studies Investigating Firm Specific Factors in Asset Pricing 8

1.2.2.1 Studies Investigating the Effects of One Firm Specific Factor on Stock Returns 8

1.2.2.1.1 Price to Earnings Ratio 8

1.2.2.1.2 Size Effect 9

1.2.2.1.3 Book to Market Ratio 11

1.2.2.1.4 Debt to Equity 12

1.2.2.2 Studies Investigating the Effects of Two or More Firm Specific Factors on Stock Returns 12

1.2.2.2.1 Studies for Developed Markets 12

1.2.2.2.2 Studies for Emerging Markets 26

(9)

viii

1.3 CAPITAL ASSET PRICING MODEL WITH HIGHER ORDER

CO-MOMENTS 33

CHAPTER TWO EMPIRICAL ANALYSIS 2.1 DATA AND THEIR SOURCES 48

2.1.1 Data Sampling Criteria 48

2.1.2 Monthly Price and Return Data of Stocks Traded in the ISE 49

2.1.3 The Market Portfolio and Risk Free Rate 50

2.1.4 Portfolio Formation 50

2.1.4.1 Size Portfolios 51

2.1.4.2 Industry Portfolios 51

2.1.4.3 Fama French Portfolios 52

2.2 PRELIMINARY ANALYSIS 53

2.3 TIME SERIES ANALYSIS 85

2.4 CROSS SECTIONAL ANALYSIS 91

CONCLUSION 96 REFERENCES 99 APPENDICES 117 APPENDIX A 118 APPENDIX B 128 APPENDIX C 133 APPENDIX D 135 APPENDIX E 136 APPENDIX F 137 APPENDIX G 138

(10)

ix

ABBREVIATIONS

AMEX American Stock Exchange

APT Arbitrage Pricing Theory

BH Portfolio contains the stocks in the big size group that are also in the high book to market- value- group

BL Portfolio contains the stocks in the big size group that are also in the

low book to market- value- group

BM Portfolio contains the stocks in the big size group that are also in the medium book to market- value- group

BE/ME Book to Market Ratio

BLEV Book Leverage

CAPM Capital Asset Pricing Model

CRSP Center for Research in Security Prices

D/(D + BE) Debt to Book Value of Total Assets D/(D + ME) Debt to Total Assets

E/P Earning to Price Ratio

EAFE Europe, Australia, and the Far East

EI/BE Earnings on Book Equity

FF Fama French Factors

FIML Full Information Maximum Likelihood

FTSE Financial Times Stock Exchange

GIL Turkish Government Internal Loan Index

GRS Gibbons, Ross and Shanken

HML High Minus Low

HMLEP High Minus Low E/P Ratio Stocks

IFC International Finance Corporation

ISE Istanbul Stock Exchange

LS Least Squares

LSE London Stock Exchange

LTS Least Trimmed Squares

(11)

x

MLEV Market Leverage

NASDAQ National Association of Securities Dealers Automated Quotations NYSE New York Stock Exchange

OLS Ordinary Least Squares

P/E Price to Earning Ratio

REIT Real Estate Investment Trust

RMSE Root Mean Squared in-Sample Pricing Error

S- _{Portfolio of stocks with the most negative coskewness}

SH Portfolio contains the stocks in the small size group that are also in the

high book to market- value- group

SL Portfolio contains the stocks in the small size group that are also in the

low book to market- value- group

SM Portfolio contains the stocks in the small size group that are also in the medium book to market- value- group

SKS Coskewness mimicking portfolio

SMB Small Minus Big

SUE Standardized Unexpected Earnings

(12)

xi

LIST OF TABLES

Table 1: Number of Stocks Included in the Analysis for Each Year p. 49 Table 2: Descriptive Statistics of Industry Portfolios p. 56 Table 3: Descriptive Statistics of Size Portfolios p. 57 Table 4: Descriptive Statistics of Fama French Portfolios p. 58 Table 5: Descriptive Statistics for the Period from July 1998 to June 2004 p. 77 Table 6: Descriptive Statistics for the Period from July 1999 to June 2005 p. 78 Table 7: Descriptive Statistics for the Period from July 2000 to June 2006 p. 79 Table 8: Descriptive Statistics for the Period from July 2001 to June 2007 p. 80 Table 9: Descriptive Statistics for the Period from July 2002 to June 2008 p. 81 Table 10: Descriptive Statistics for the Period from July 2003 to June 2009 p. 82 Table 11: Descriptive Statistics for the Period from July 2004 to June 2010 p. 83 Table 12: Results of the Multivariate Test of Gibbons–Ross–Shanken (1989) p. 90 Table 13: Cross Sectional Analysis Results p. 95 Table 14: Some of the Important Empirical Studies about the CAPM and its

Versions p. 118

Table 15: Studies about Multifactor Models for the Developed Markets p. 128 Table 16: Studies about Multifactor Models for the Emerging Markets and

Turkey p. 133

Table 17: Studies about Higher Moments in Asset Prices p. 135 Table 18: ISE 100 Index Monthly Return over the Analysis Period p. 136 Table 19: Monthly Risk Free Rate over the Analysis Period p. 137 Table 20: Monthly Excess Returns of Industry Portfolios p. 138 Table 21: Monthly Excess Returns of Size Portfolios p. 141 Table 22: Excess Returns of Fama French Portfolios p. 144 Table 23: Excess Returns of Momentum (6,1) Portfolio p. 147 Table 24: Excess Returns of Momentum (6,6) Portfolio p. 150 Table 25: Excess Returns of Momentum (6,12) Portfolio p. 153

(13)

xii

LIST OF FIGURES

Figure 1: Twelve-Month Moving Averages of Skewness for Industry 1 and Market

Portfolio p. 60

Portfolio p. 61

Portfolio p. 62

Portfolio p. 63

Portfolio p. 64

Figure 10: Twelve-Month Moving Averages of Skewness for Industry 10 and

Market Portfolio p. 64

Figure 11: Twelve-Month Moving Averages of Skewness for Size 1 and Market

Portfolio p. 65

Portfolio p. 66

(14)

xiii

Portfolio p. 67

Portfolio p. 68

Portfolio p. 69

Figure 21: Twelve-Month Moving Averages of Skewness for SH and Market

Portfolio p. 70

Figure 22: Twelve-Month Moving Averages of Skewness for SM and Market

Portfolio p. 70

Figure 23: Twelve-Month Moving Averages of Skewness for SL and Market

Portfolio p. 71

Figure 24: Twelve-Month Moving Averages of Skewness for BH and Market

Portfolio p. 71

Figure 25: Twelve-Month Moving Averages of Skewness for BM and Market

Portfolio p. 72

Figure 26: Twelve-Month Moving Averages of Skewness for BL and Market

(15)

1

INTRODUCTION

Asset pricing that has more than three hundred years old history (Cheng and Tong, 2008:1) is always an attractive topic for investors and finance scholars. The Capital Asset Pricing Model (CAPM) of William Sharpe (1964) and John Lintner (1965) marks the birth of asset pricing theory (Fama and French, 2004:25). The CAPM has attracted a great deal of attention by the finance world since it was developed. It is still widely used in academic research and applications, such as portfolio management and estimating the cost of capital.

The CAPM’s main implication is that the variation of excess stock returns can be explained by the market factor –systematic risk- alone. During the last half- century, the validity of the model has been investigated. Even though early studies about the CAPM have found some supporting evidence (Black, Jensen and Scholes (1972), Fama and MacBeth (1973)), a large body of literature has documented the invalidity of the model

The CAPM has taken a large number of criticism because of its unrealistic assumptions. From the time of the creation of the model to the present, some versions of the model have emerged. In recent years, the discussion is focused on the two assumptions of classical CAPM: asset returns have normal distribution and market excess return is the unique factor in explaining the variation of excess returns. In the literature, there have been studies providing supporting evidence against both of these assumptions. Some studies have concluded that asset returns are

characterized by skewness and significant leptokurtosis1; providing evidence against

the normality assumption. Again, some other studies have found ignored firm

specific factors such as size and value as significant factors2. Up to the study of

1

Examples of evidence on the nonnormality of returns include Chunhachinda et al. (1997) that represent the returns of the world's 14 major stock markets are not normally distributed. For ISE, Harris and Kucukozmen (2001) reject the normality of daily equity returns.

2

On the second criticism area, there have been a large number of studies identifying a number of variables that have explanatory power on the cross sectional variation in stock returns in addition to

(16)

2 Harvey and Siddique (2000), these assumptions were analysed seperately. Harvey and Siddique (2000) take into consideration the two criticized points together and develop “Four Factor Model” by adding coskewness factor to the Three Factor Model that includes size and value premium in addition to market beta. The study is taken as a basis for most subsequent studies considering skewness in the asset pricing models.

Lin and Wang (2003) adopt the methodology of Harvey and Siddique (2000) for Taiwan stock market during the period January 1986 to December 2000. For Istanbul Stock Exchange (ISE), the study of Mısırlı and Alper (2008) is the first and only study investigating the relative importance of coskewness in explaining the variation of excess returns in ISE. They investigate the impact of coskewness on the variation of portfolio excess returns in ISE over the period July 1999 to December 2005 by adopting similar methodology with Harvey and Siddique (2000).

In this dissertation, the CAPM, the Fama French Three Factor model and the Four Factor Model are implemented. The purpose of this dissertation is to investigate the impact of coskewness on the variation of portfolios that are formed according to industry, size, book to market ratio and momentum for ISE over the period July 2002 to June 2010 by using the similar methodology to Harvey and Siddique (2000).

This dissertation adds to the existing literature on ISE and emerging markets in two dimensions. There is no previous dissertation trying to reveal the possible incremental effect of coskewness factors over market factor and size and value factors in explaining the variations of excess returns in ISE. The other intended contribution of this dissertation is to use an extended dataset compared to previous

the market beta. Firm size (Banz (1981), Keim (1983)) and book-to-market equity (BE/ME) (Rosenberg et al., 1985; Chan, Hamao and Lakonishok (1991) are two of the most significant factors in explaining the cross-section of average returns. Aydogan and Gursoy (2000), Gonenc and Karan (2003), Bildik and Gulay (2002), Aksu and Onder (2003), Gokgoz (2007) are some of the important researches investigating the effects of firm specific factors on the stock returns of ISE firms.

(17)

3 studies on ISE. The dataset includes all stocks that are listed in ISE during 1998 - 2010.

The remainder of this dissertation is organized as follows. In chapter one, a brief history of the CAPM, multifactor models and asset pricing models with higher order co-moments and examination of the previous studies that fall in to their categories will be given. In the first section of chapter two, the data and their sources will be presented. Then, data sampling criteria and portfolio formation process will be explained and preliminary statistics will be given. In the next section, the effectiveness of the CAPM, the Three Factor Model and the Four Factor Model will be tested and compared through time series analysis. In this section, the multivariate test of Gibbons, Ross and Shanken (GRS) (1989) will be applied to ISE. In the last section of this chapter, cross sectional regressions will be run following Fama– MacBeth (1973) as well as full information maximum likelihood (FIML) to investigate the incremental power of coskewness over CAPM and Fama–French factors. In the last section, conclusion part will be introduced.

CHAPTER ONE ASSET PRICING MODELS 1.1 THE CAPITAL ASSET PRICING MODEL

The CAPM developed by Sharpe (1964) and Lintner (1965) relates the expected rate of return of an individual security to its systematic risk. The foundations for the development of asset pricing models were laid by Markowitz (1952) and Tobin (1958) (Galagedera, 2007:821). Markowitz (1952) argues for the explicit recognition of risk and its quantification in terms of variance. He also introduces the notion of a mean-variance efficient portfolio as one that (1) provides minimum variance for a given expected return and (2) provides maximum expectereturn for a given variance (Markowitz, Todd and Sharpe, 1987:x). His model represents the efficient frontier of portfolios and the investors are expected to select a portfolio in accordance with their risk preferences from the efficient portfolio set.

(18)

4 According to the model of Markowitz, investors only care about the mean and variance of portfolio alternatives during one-period investment horizon. Sharpe (1964) and Lintner (1965) add two key assumptions to the Markowitz model to identify a portfolio that must be mean-variance-efficient. The first assumption is complete agreement, and second is that there is borrowing and lending at a risk -free rate, which is the same for all investors and does not depend on the amount borrowed or lent. (Fama and French, 2004:26).

Tobin (1958) shows that under certain conditions Markowitz’s model implies that the process of investment choice can be broken down into two phases: first, the choice of a unique optimum combination of risky assets; and second, a seperate choice concerning the allocation of funds between such a combination and a single riskless asset (Sharpe, 1964:426).

These theories were later expanded by Treynor (1961), Sharpe (1964), Lintner (1965) and Mossin (1966) (Rejepova, 2005:3). They develop a computationally efficient method, the single index model, the CAPM, where return on an individual security is related to the return on a common index (Jones, 1991). The notation that the CAPM bases on is that intelligent, risk –averse shareholders will seek to diversify their risks, and, as a consequence, the only risk that will be rewarded with a risk premium will be a systematic risk (Shapiro and Sarin, 2009:415).

The CAPM is developed under these assumptions (Focardi and Fabozzi, 2004:512):

x Investors make investment decisions based on the expected return and

variance of the return.

x Investors are rational and risk averse.

x Investors subscribe to the Markowitz method of portfolio

diversification.

(19)

5

x Investors have the same expectations about the expected return and

variance of all assets.

x There is a risk-free asset and investors can borrow and lend any

amount at the risk-free rate.

x Capital markets are completely competitive and frictionless.

The assumptions of the CAPM imply that the market portfolio M must be on the minimum variance frontier if the asset market is to clear. This means that the algebraic relation that holds for any minimum variance portfolio must hold for the market portfolio. Specifically, if there are N risky assets (Fama and French, 2004:28);

(Minimum Variance Condition for M) E(Ri) = E(RZM) + [E(RM) - E(RZM)]βiM, i = 1, .. , N.

where E(Ri) is the expected return on asset i, and βiM, the market beta of asset i, Cov (Ri, RM) is the covariance of its return with the market return divided by the variance of the market return,

Another implication of the CAPM is that all investors hold the market portfolio and non-systematic risk is diversified away. So, the only systematic risk is taken into account.

The CAPM predicts that the expected return on an asset above the risk -free rate is linearly related to the systematic risk, which is measured by the asset’s beta. This model assumes that, investors only care about expected return and variance of the return in their investment decision making process and are exposed to only systematic risk that is represented by “beta” in this model equation. An asset can have positive, negative or zero beta. If an asset has positive beta, it moves in the

) R ( ) R , R ( Cov M 2 M i M i V E

(20)

6 same direction as the market portfolio. For the assets having a negative correlation to the market portfolio, the beta will be negative and they will move in the opposite direction to the market portfolio. On the other hand, beta of the assets having zero correlation between their returns and market’s return such as risk free assets are zero.

During the last half century, there have been numerous studies investigating the validity of the CAPM for developed and emerging markets. In early studies about the CAPM, some supporting evidence about the model have been found (Black, Jensen and Scholes (1972), Fama and MacBeth (1973)). But by the early 1970s, several studies suggest that there are deviations from the linear CAPM risk return trade-off due to other variables that affect this tradeoff. The purpose of these studies are to find the components that CAPM is missing in explaining the risk -return trade-off and to identify the variables that create those deviations (Michailidis et al., 2006:80). As a result of these studies, modifications to the model have been made and versions of the model have emerged (Lintner (1971), Sharpe and Cooper (1972), Mayers (1976), Merton (1973), Rubinstein (1976), Elton and Gruber (1978), Breeden, Gibbons and Litzenberger (1989), Fama and French (1995), Kraus and Litzenberger (1976), Harvey and Siddique (2000)). Zero Beta CAPM, Consumption Based CAPM, Multifactor Models, CAPM with Higher Order Co-moments, Multi-Period CAPM and Multi-Beta CAPM are the most widely used versions of the CAPM. There are numerous studies conducted about the traditional CAPM and its versions. In Appendix-A, some of the important empirical studies about the CAPM and its versions are given.

1.2 MULTI-FACTOR MODELS

The majority of the asset pricing literature provide evidence that the variation in the excess stock returns cannot be explained by the market beta alone and show that macroeconomic factors and some firm specific factors such as size and book to market ratio (BE/ME) can explain a sizeable portion of these variations.

(21)

7 Multi factor asset pricing models are constructed by adding these significant factors and various macro economic factors to the market factor in the CAPM. In this section, the Arbitrage Pricing Theory (APT) that considers macro economic factors and the studies investigating firm specific factors in the asset pricing concept will be addressed. The Fama French Three Factor Model, regarded as the most important of the studies that investigate firm specific factors, will be described in detail.

1.2.1 The Arbitrage Pricing Theory

The Arbitrage Pricing Theory is a multi-factor model developed by Ross (1976). The APT assumes that stock returns are generated according to factor models. According to the APT, there are n systematic factors that cause asset returns to systematically deviate from their expected values. The theory does not specify how large the number n is, nor does it identify the factors. It simply assumes that these n factors cause returns to vary together. There may be other, firm-specific reasons for returns to differ from their expected values, but these firm-specific deviations are not related across stocks. Since the firm-specific deviations are not related to one another, all return variation not related to the n common factors can be diversified away (Davis, 2001:3).

The equation of the APT is given below:

E(rj) = rf + bj1RP1 + bj2RP2 + … + bjnRPn i=1, …, N Ei= ith asset’s expected return

rf = risk free rate

bi1= sensitivity of the ith asset to factor n also called factor loading, RPk= risk premium of the factor

The equation states that the security’s expected return is related to the security’s factor betas. Each factor represents risk that cannot be diversified away. The higher a security’s beta with regard to a particular factor is, the higher is the risk that the security bears. In a rational world, the expected return on the security should

(22)

8 compensate for this risk. The above equation states that the expected return is a summation of the risk-free rate plus the compensation for each type of risk that the security bears (Ross, Westerfield and Jaffe, 2002:299).

1.2.2 Studies Investigating Firm Specific Factors in Asset Pricing

The APT studied the effects of a few pervasive factors that are dominant source of co-variation among asset returns. While the studies continue to be made to enhance the CAPM, towards the 1980s, firm specific factors are started to evaluate in asset pricing concept. There have been lots of empirical studies investigating those factors in this concept. Among the studies, the one that had created a tremendous impression and widely used in the finance world is the Fama French Three Factor Model.

This sub-section is divided into two parts. In the first part, the previous important studies that only considers one firm specific factor -size, book to market, price to earnings (P/E) ratio and debt to equity -in addition to the market factor will be addressed. In the second part, the previous important studies analyzing the effects of two or more firm specific factors jointly on the stock returns, again in addition to the market factor will be addressed.

1.2.2.1 Studies Investigating the Effects of One Firm Specific Factor on Stock Returns

1.2.2.1.1 Price to Earnings Ratio

One important factor that is used to explain the variation of the excess stock returns is price to earning ratio. Price to earnings ratio displays the relative stock returns when other factors are kept constant. It is widely argued that the P/E ratio implicitly incorporates the perceived riskiness of a given company's future earnings and it reflects the structure of the balance sheet of the company (Setiawan,2010:24).

(23)

9 Basu (1977) empirically tests whether the investment performance of common stocks is related to their P/E ratios by using the database that represents over 1400 industrial firms, which actually traded on the New York Stock Exchange (NYSE) between September 1956-August 1971. He computes the P/E ratio of every sample security and rank them in an ascending order. Then he divides the securities into five portfolios in accordance with this ranking. He implements the CAPM to the excess returns of these portfolios. According to the result of this study, the low P/E portfolio groups have earned higher absolute and risk-adjusted rates of return than the high P/E securities. This result asserts that P/E ratio is related to average stock returns in the U.S.

P/E ratio is used in later studies with other firm specific factors, such as size,

book to market ratio etc. 3These studies will be addressed under the heading of

Studies Investigating The Effects of Two or More Firm Specific Factors On Stock Returns.

1.2.2.1.2 Size Effect

As mentioned before, firm specific factors’ explanatory effects on the variation of the stock returns are studied starting from the years of 1980s. While in the earlier studies, the additional explanatory power of size factor over the CAPM is investigated, subsequent studies use this factor with other firm specific factors, especially with book to market ratio.

Banz (1981) investigates whether the size effect exists as a relevant factor for asset pricing. He examines the relationship empirically between the return and the total market value of NYSE common stocks during the period 1936-1975. He uses the following asset pricing model in this study:

3

e.g. Reinganum (1981), Basu (1983) and Jaffe, Keim and Westerfield (1989) examine the size and earnings- price factors’ effects jointly on the variation of stock returns. Chan, Hamao and Lakonishok (1991) uses earnings yield, size, book to market ratio, and cash flow yield for Japanese stocks.

(24)

10

E(R

i

) = γ

0

+ γ

1

β

1

+ γ

2

[(φ

i

- φ

m

)/ φ

m

]

where

E(Ri): Expected return on security i,

γ

0: Expected return on a zero beta portfolio

γ

1: Expected market risk premium

γ

2: Constant measuring the contribution of

I

i to the expected return of a

security

φ

i:Market value of security i and

φ

m: Average market value

He finds that on average, small NYSE firms have significantly larger risk over the whole analysis period. Investors can earn from holding very small firms long and very large firms short, on average, 1,52 percent per month or 19,8 percent on an annualized basis. The size effect is not linear in the market proportion but it is most pronounced for the smallest firms in the sample.

Keim (1983) examines the empirical relation between abnormal returns and market value of NYSE and American Stock Exchange (AMEX) common stocks month by month over the period 1963-1979. His findings show that daily abnormal return distributions in January have large means relative to the remaining eleven months, and that the relation between abnormal returns and size is always negative and more pronounced in January than in any other month.

Chan and Chen (1991) examine the differences in structural characteristics that lead firms of different sizes to react differently to the same economic news. They find that a small firm portfolio has low production efficiency and high financial leverage. In their analysis they construct two size-matched return indices designed to mimic the return behavior of marginal firms and find that these return indices are important in explaining the time-series return difference between small and large

(25)

11 firms. They conclude that the size effect is mainly driven by marginal firms in distress.

1.2.2.1.3 Book to Market Ratio

Another important and one of the most used firm specific factors in asset pricing is book to market ratio. According to the studies investigating developed markets, small capitalization, high BE/ME “value” stocks earn higher returns than high priced “growth” stocks.

Rosenberg, Reid and Lanstein (1985) examine the relationship between stock returns and book-to-market ratio for the common stocks of NYSE firms during the period January 1973- September 1984. They find that firms with high BE/ME have higher returns than firms with low BE/ME (Drew and Veeraraghavan, 2002:337-338).

Lakonishok, Schleifer and Vishny (1994) try to make potential statements for why portfolios with high BE/ME (value strategies) outperform the market, using the stocks in the NYSE and AMEX during the period April 1963- April 1990. They bring an explanation about why value strategies have produced superior returns by using the overreaction hypothesis, like the studies of DeBondt and Thaler (1987) and Haugen (1995).

Their results establish three propositions. First, a variety of investment strategies that involve buying out of favor stocks have outperformed glamour strategies over the analysis period. Second, the market participants appear to have consistently overestimated future growth rates of glamour stocks relative to value stocks. Third, using conventional approaches to fundamental risk, value strategies appear to be no riskier than glamour strategies.

(26)

12

1.2.2.1.4 Debt to Equity

Researchers found that leverage is another factor affecting the stock returns. Bhandari (1988) investigates the effect of debt to equity ratio on the expected common stocks returns during the period 1948- 1981. He divides the difference between the book value of total assets and common equity to the market value of common equity and uses as debt to equity ratio. According to his result, expected common stock returns are positively related to the ratio of debt (non-common equity liabilities) to equity, controlling for the beta and firm size and including as well as excluding January, though the relation is much larger in January.

1.2.2.2 Studies Investigating the Effects of Two or More Firm Specific Factors on Stock Returns

The mentioned previous studies above that examine the effects of only one firm specific factor as an additional factor to the market factor in explaining the variation of excess stock returns, present evidence against the CAPM. Those additional factors such as size, book to market ratio etc. have been shown to be significant determinants of variation of excess returns. From this point forth, studies that incoporate two or more firm specific factors into the CAPM have been made. In this part, those studies for developed and emerging markets will be addressed under the two headings.

1.2.2.2.1 Studies for Developed Markets

Reinganum (1981), Basu (1983) and Jaffe, Keim and Westerfield (1989) examine the size and earnings- price factors’ effects jointly on the variation of stock returns. Reinganum (1981) tests the validity of the CAPM and efficient market hypothesis. He uses the model of Latane, Jones and Rieke (1974) and Latane and Jones (1977) about standardized unexpected earnings (SUE) to test whether abnormal returns can be earned during the period from 4th quarter of 1975 to 3rd quarter of 1977. He constructs low and high SUE portfolios and test whether the difference in the expected returns between these portfolios is zero. The mean

(27)

13 difference between these portfolios is not found statistically significant. The result supports the assumption of market efficiency and the CAPM. Then he constructs ten earnings to price ratio (E/P) and intersection of size and E/P portfolios by using the ranked E/P and market value of analysed firms by using the annual data from 1963 to 1977. He implements the CAPM for these constructed portfolios and compares mean excess return, beta and average median of them. His findings suggest that the simple one-period capital asset pricing model is misspecified. The set of factors omitted from the equilibrium pricing mechanism seems to be more closely related to firm size than E/P ratios.

Basu (1983) examines the empirical relationship between earnings' yield, firm size and returns on the common stock of NYSE firms during the period 1963-1979. He shows that E/P is a significant factor in explaining the cross-section of average returns on U.S. stocks and there is a positive relationship between average return and E/P during the analysis period. Also he claims that E/P effect subsumes the size effect due to the disappearance of size effect when both variables are jointly considered (Jaffe, Keim and Westerfield 1989:135).

Jaffe, Keim and Westerfield (1989) reexamine the relation between stock returns and the effects of size and earnings to price ratio for the period 1951- 1986 – longer sample period than the studies previously made for the two factor- by considering the difference between January and other months. Different findings and opinions about the relationship and interaction between size and E/P factors motivate them to make this research.

They use two ranking procedures in the portfolio formation. In the first one, initially they rank the analysed firms on the ratio of year-end earnings to share price at the end of March in each year and placed into one of six groups. Then, they rank the stocks in each E/P group on the March 31 market value of their common stock outstanding. The procedure in the second one is identical to the first one except that firms are ranked on market value first and then ranked on E/P.

(28)

14 They use the following Seemingly Unrelated Regression (SUR) model to test for significance of the size and E/P effects by considering the differences in the effects in January versus the other months:

Rpt- RFt= α0j Djt + α0r(1- Djt) + βpj(Rmt-RFt)Djt + βpr(Rmt-RFt) (1- Djt) + α1j [(E/Ppt)Djt] + α1r [(E/Ppt)Djt] + α2j (LMVEptDjt) + α2r [(LMVEpt (1-Djt)] + ept

p=1,…,25 and t=1,…,T

where Djt is a dummy variable that takes the value of one if month t is January and zero otherwise, Rmt is the monthly return for the market index, RF is the monthly return on a risk free asset, E/Ppt is the average earnings to price ratio of the

securities in portfolio p for time t, and LMVEpt is the natural logarithm of the

average market value of outstanding common stock in the portfolio for time t.

According to their findings, E/P and size effects are significant for the whole sample period. The result is inconsistent with Basu (1983) and Reinganum (1981). When they consider the January effect, they find a difference between January and the rest of the year; the coefficients on both E/P and size are significant in January, but only the E/P coefficient is significant outside of January and their portfolio formation procedures do not affect the results on E/P. They also find evidence of consistently high returns for firms of all sizes with negative earnings.

Chan, Hamao and Lakonishok (1991) examine the cross-sectional differences in returns on Japanese stocks to the underlying behavior of four fundamental variables: earnings yield, size, book to market ratio, and cash flow yield. They find that there is a significant relationship between these fundamental variables and expected returns in the Japanese market. The book to market ratio and cash flow yield have the most significant positive impact on expected returns among the four fundamental variables.

(29)

15 Fama and French (1992) investigate the roles of market beta, size, leverage, book to market equity and earnings-price ratios on the cross sectional variation in average stock returns. They find that beta does not seem to help explain the cross-section of average stock returns and the combination of size and book to market equity seems to absorb the roles of leverage and E/P in average stock returns during their 1963-1990 sample period.

Coming to the year 1993, Fama and French (1993) extend the study of Fama and French (1992) and the worldwide known Fama French Three Factor Model is constructed. This model has made tremendous impact on the financial world. In this study they consider bond returns in addition to common stock returns during the period 1963- 1990 as a distinction from their study at 1992. They expand the set of variables used to explain returns of bond returns to examine whether variables that are important in bond returns help to explain stock returns and vice versa. Also they change their approach in testing asset pricing models. They use time series regression approach of Black, Jensen, and Scholes (1972) instead of cross sectional analysis.

In time series analysis, the explanatory variables used fall into two sets, those are for explaining the variation of bond returns and those are for stocks. The unexpected changes in interest rates and default risk factor are used as explanatory variables for bond returns. As for stock returns, they use the variables that mimic the risk factors in returns related to size and value of the firms.

Fama and French (1992) use six portfolios formed from sorts of stocks on market capitalization and book to market equity. They use the same six portfolios in this study to form the intersections of size and book to market equity portfolios. In June of each year during the whole analysis period, they simple sort of all NYSE stocks on Center for Research in Security Prices (CRSP) into two groups on market capitalization. They use the median NYSE size to split NYSE, AMEX and National Association of Securities Dealers Automated Quotations (NASDAQ) stocks. These groups are called by small and big (S and B).

(30)

16 They also break these stocks into three book to market equity groups based on the ranking of them on their book to market values for each year. They call the top of 30% of the ranked values of BE/ME is High (H), middle 40% is Medium (M) and bottom 30% Low (L). They compute book to market equity of each stock by dividing the book common equity for the fiscal year ending in calendar year t-1 to market equity at the end of December of t-1. Then, they construct size portfolios of SL, SM, SH, BL, BM and BH from the intersections of two size portfolios and the three BE/ME groups. For example SH portfolio contains the stocks in the small size group that are also in the high book to market- value- group.

They choose stocks to be in a portfolio group by considering some criteria.

x They do not use negative BE/ME firms.

x They use only firms with ordinary common equity as classified by

CRSP.

x Portfolios are formed for the period from July of year t to June of

t+1 and reformed in June of t+1.

x A firm must have CRSP stock prices for December of year t - 1 and

June of t and COMPUSTAT book common equity for year t – 1 to be included in the tests.

x They do not include firms until they have appeared on

COMPUSTAT for two years to avoid the survival bias inherent in the way COMPUSTAT adds firms to its tapes

They calculate monthly value- weighted returns on the six portfolios by considering these criteria.

As for size factor they construct a portfolio SMB –small minus big- that mimics the risk factor in returns related to size. They compute the monthly return of this portfolio by using this equation:

(31)

17 As for value factor they construct a portfolio HML –high minus low- that mimics the risk factor in returns related to value. They compute the monthly return of this portfolio by using this equation:

HML = ½ (SH + BH) – ½ (SL + BL)

They construct twenty five intersection portfolios by using the same portfolio formation procedure in the six intersection portfolios and examine the relationship of their excess returns and explanatory variables. They first examine the explanatory power of bond and stock market factors separately to test whether bond- market factors that are important in bond returns capture common variation in stock returns and vice versa. Then, they examine the joint explanatory power of the bond and stock-market factors, to develop an overall story for the common variation in returns. Because of my dissertation is focused on stock market, the sections of this study about stock market factors is reviewed.

In stock-market factors analysis, they examine three regressions. In the first one, they use only market excess return as an explanatory variable. SMB and HML are used jointly in the second regression. In the third regression all variables - market excess return, SMB and HML- are used as explanatory variables. This regression model is called as the Three Factor Model that considers market, size and value factors in explaining the variations of stocks and bonds returns.

R

i

– R

f

= α

i

+ β

i

(R

m

– R

f

) + s

i

(SMB) + h

i

(HML)

where Ri-Rf is the excess return of each intersection portfolio, Rm is the

value-weighted percent monthly return on all the stocks in the analysis. Rf is the one-month Treasury bill rate, observed at the beginning of the one-month. SMB (small minus big) is the return on the mimicking portfolio for the size factor in stock returns. HML (high minus low) is the return on the mimicking portfolio for the book -to-market factor.

(32)

18 Their findings are consistent with the study of Fama and French (1992). They suggest that there are common return factors related to size and book -to-market equity that help capture the cross-section of average stock returns in a way that is consistent with multifactor asset-pricing models.

Claessens, Dasgupta and Glen (1995) examine the cross-sectional pattern of returns in 19 developing countries empirically. They use data compiled by the International Finance Corporation (IFC) for analyzed countries during the period of 1986-1993, which provides 96 monthly observations for each country. They use size, trading volume, dividend yield, earnings/ price and cross exchange rates in addition to beta as explanatory variables in their cross sectional model. According to their results, in addition to beta, two factors, size and trading volume have significant explanatory power in a number of these markets; dividend yield and earnings/price ratios are also important, but in slightly fewer markets. For Turkey, size, E/P, dividend yield and turnover that express the value traded measured in dollars relative to the number of shares outstanding are found as significant factors.

Fama and French (1995) study whether the behavior of stock prices, in relation to size and book-to market-equity, is consistent with the behavior of earnings to provide an economic foundation.

They form six size- book to market intersection portfolios by using the same procedure and databases of Fama French (1993) for each year during the period from 1963 to 1992. As a measure of profitability, they use earnings on book equity that is the ratio of common equity income for the fiscal year ending in calendar year t to the book value of common equity for year t - 1.

They draw a plot representing average profitability as a function of size and BE/ME for a long period around the portfolio formation. Their aim is to reveal how earnings behave before firms are classified as small or big on ME and low or high on BE/ME, and how profitability evolves in the years after portfolio formation.

(33)

19 Their other research questions are, whether there are market, size and BE/ME factors in earnings shocks like those in stock returns, and whether variation in stock returns traces to common factors in stock returns. Firstly they adopt Fama French Three Factor Model to examine the relationship between market, size and BE/ME factors in stock returns. Then they investigate the relationship between the change in

fundamental variables (EI/BE, ln EBI, and lnS)4 and market, size and BE/ME proxies

for the analysis period. They express this relationship as follows:

) 1 t ( e ) 1 t ( HML h ) 1 t ( SMB s ) 1 t ( Mkt b a ) 1 t ( Y ' ' ' '

where 'Y(t1) is the change in a fundamental variable (EI/BE, ln EBI, or ln

S) from year t to t + 1 for all firms in each six intersection portfolio.

'

Mkt

,

'

SMB

and HML' are the market, size and BE/ME factors in Y' .

Their results indicate that, as they expected, firms with high BE/ME have low ratios of earnings to book equity and vice versa. Small stocks within book-to-market groups tend to be less profitable than big stocks. According to the results of Fama French Three factor model for market, size and book to market factors in stock returns and in earnings and sales, there are market, size and book to market factors in fundamentals that are similar to those in stock returns. The market and size factors in earnings help explain the market and size factors in returns, but it is not valid for book to market factor.

Barber and Lyon (1997) adopt the methodology of Fama French (1992) for both financial and nonfinancial firms. Fama and French (1992) exclude financial firms from their analysis, because of their different leverage characteristics. Barber and Lyon (1997) want to investigate whether the relationship between firm size, book-to-market ratios, and security returns for financial and nonfinancial firms is similar or not during the period July 1973-December 1994. They use the firms with available returns data on the CRSP NYSE/AMEX/NASDAQ monthly returns files

4

EI/BE: Equity income/book equity, EBI: Equity income plus interest expense and preferred dividends summed over all stocks in a portfolio, S is the sum of sales for the stocks in a portfolio.

(34)

20 from July 1973 through December 1994 in their analysis. As a result, they find a similar relation between firm size, book-to-market ratios, and security returns for financial and nonfinancial firms. In addition, they present evidence that survivorship bias does not significantly affect the estimated size or book-to-market premiums in returns.

After the studies of Fama and French (1992 and 1993), there have been large number of studies to investigate their findings robustness. Knez and Ready (1997) investigate the robustness of the estimated risk premia for size and book -to-market factors by using a new regression technique, called least trimmed squares (LTS). This technique obtained by incorporating a robust regression technique into the Fama MacBeth procedure, is not sensitive to outliers or leverage points. This robust technique allows them to isolate influential observations if the estimates are driven by a small subset of firms or months. Also, they use the least squares (LS) technique and compare the results to those obtained using LTS. Their aim is to reveal whether the size and value effect are driven by extreme observations. They use the same data as in Fama and French (1992). According to their results, the negative relation between firm size and average returns is driven by a few extreme positive returns in each month. In fact, when only 1 percent of each month's observations are trimmed, there is a significant positive relation between firm size and average returns.

Fama and French (1998) examines whether there is a value premium in markets outside the United States and whether a risk model that describes the U.S. returns can describe an other market returns. They study returns on market, value, and growth portfolios for the U.S. and twelve major EAFE (Europe, Australia, and the Far East) countries.

They construct value and growth portfolios for analyzed markets during the period of 1975 to 1995. Then they implement the CAPM and two factor CAPM that includes the difference between the global high and low book to market returns in addition to global market excess return for these portfolios. Their results indicate that value stocks tend to have higher returns than growth stocks in twelve of thirteen

(35)

21 major markets during the analysis period. An international CAPM cannot explain the value premium in international returns. But a two factor CAPM that explains returns with the global market return and a risk factor for relative distress captures the value premium in country and global returns.

Clare, Priestley and Thomas (1998) explore the relationship between beta and expected returns by using a one-step estimator to McElroy, Burmeister and Wall (1985) in addition to two step estimator due to Fama and MacBeth (1973) algorithm. They test for a linear and positive relationship between beta and expected returns for a sample of UK stock returns over the period 1980-1993. Until this study, all studies of the relationship between beta and expected stock returns have been used US stock return data.

They integrate the Fama French explanatory variables by estimating a series of cross-sectional regressions using the CAPM errors as dependent variables and the accounting variables (appropriately lagged) as independent regressors. When they estimate the CAPM by using one step approach, they find that beta has a significant and powerful role and Fama and French variables do not have any significant role for in explaining expected returns in contrast to US findings. When they consider Fama MacBeth t statistics, price effect is found as a significant factor in the UK stock market.

Chen and Zhang (1998) compare the return experience of value stocks across the six countries. The countries are the well-established market of the United States; is less persistent for the growth markets of Japan, Hong Kong, and Malaysia; and is almost nonexistent for the high growth markets of Taiwan and Thailand. Their findings reveal that strong value stock effects persist in the United States, are less persistent in Japan, Hong Kong, and Malaysia; and are undetectable in Taiwan and Thailand.

Trecartin (2000) examines whether the book equity-to-market equity ratio and other value/ growth variables predict returns consistently from 1963 to 1997 using

(36)

22 monthly intervals. He examines the reliability of the book to market ratio by comparing its relation with the return in ten year periods and five year sub-periods. He uses all firms on NYSE, AMEX, and NASDAQ if they meet the used criteria. He uses the criteria mentioned in the study of Fama and French (1993) and additional two criteria: A firm must have sales in at least two adjacent years during the five years preceding year t in order to calculate sales growth rates, and the firm must also record earning. Stock return is used as the dependent variable in monthly regressions as in Fama and French (1992). As variables, he uses book to market ratio, cash flow, weighted growth in sales and June ending market value of firms studied.

He finds that the book-to-market effect is statistically related to return as predicted in less than 50% of the monthly time periods examined. Also, the variable is not always significant in five-year sub-periods. However in ten-year periods BE/ME is significantly related to return. Thus the data supports the view that the BE/ME variable is not a reliable predictor of return over short time horizons.

Dimson, Nagel and Quigley (2003) investigate the UK stock market by using a new dataset of accounting information that covers virtually all UK firms ever listed on the London Stock Exchange (LSE) during the period of 1995- 2001. They evaluate the size and value effects on the analyzed firms’ stock returns. Even though their focus is on book-to-market as a measure of value, they also provide some information on the role of dividend yields. They form portfolios like the mechanism of Fama and French (1993) but a few adjustments done where necessary to account for characteristics of the UK data.

They find that there is a strong value premium in the UK for the period 1955-2001. The value premium exists within the small-cap as well as the large-cap universe. They also find that dividend yield as a measure of value produces strikingly similar results.

Dunis and Reilly (2004) analyze the investment returns of best decile growth stocks and value stocks portfolios by using daily data over the period 31st December

(37)

23 2000 to 31st December 2002. They use 5 variables to categorize a panel of 689 stocks from the FTSE All-Share index. These variables are price/book value ratio, price/earnings ratio, cash flow/price ratio, dividend yield and value of firms. As firms’ value, they assume that investors have priced in their opinions of future prospects of firms and give them a value.

Their results suggest that a “value-growth” factor is significant in the UK stock market no matter which of the five relative valuation techniques are used. Value stocks outperform growth stocks, on average, for all five relative valuation techniques used during the period studied, both absolutely and after adjustment for risk. Value stocks also outperform the market, on average, for all five relative valuation techniques, both absolutely and after adjustment for risk. The lowest market capitalization decile portfolio is the best performing relative valuation technique, in terms of its risk-adjusted return, and the level of returns for all the value deciles are significant no matter which of the five methods of selecting the value decile portfolio are used.

Charitou and Constantinidis (2004) examine empirically the Fama and French three factor model of stock returns using Japanese data over the period 1991- 2001. Their main aim is to provide evidence that would contribute to the effort of explaining the Fama and French three factor model in a country that differs substantially from the US not only with regards to its financial reporting system but also as it relates to its economic characteristics. They use all industrial firms with ordinary common equity included in the Global Vantage database over the analysis period. They construct six intersection portfolios like in the study of Fama French (1993) and examine whether market equity (size, ME) and book-to-market equity (BE/ME) are related to profitability of firms when profitability is measured by earnings on book equity (EI/BE).

They find that the size and BE/ME effect are not very clear in Japan. When the testing portfolios consist of small stocks, the explanatory power of the size factor (SMB) dominates the explanatory power of the BE/ME factor (HML). The opposite

(38)

24 occurs when the testing portfolios consist of big stocks. In US the explanatory power of the HML factor always dominates the explanatory power of the SMB factor. They explain that the difference may be due to the economic crisis in Japan during the period examined, and specifically to the low profits of small low-BE/ME stocks.

Peterkort and Nielsen (2005) investigate whether the book to market ratio acts as a proxy for risk by developing a leverage-based alternative to traditional asset pricing models.

They use only ordinary common shares and firms traded on the NYSE, the AMEX, or the NASDAQ over the period 1978- 1995. They measure market leverage (MLEV) as debt to total assets D/(D + ME), and book leverage (BLEV) as debt to book value of total assets D/(D + BE). They use share price and firm size to control for sample bias problems. Transaction and liquidity costs, bias in measured returns, and sample-selection bias are all expected to be most severe in small, low-price stocks. They define size and price control variables as ln(ME) and 1/(price per share).

They demonstrate the relations among MLEV, BLEV, BE/ME, and average return by constructing portfolios. Each year they sort the base sample into 10 MLEV decile portfolios. They then sort each MLEV decile portfolio into 10 BE/ME subdecile portfolios, resulting in 100 portfolios each year. They calculate the average value for BE/ME, BLEV, and average monthly returns for each portfolio each year. They find no relation between average stock returns and the book-to-market ratio in all-equity firms after controlling for firm size, and an inverse relation between average stock returns and the book-to-market ratio in firms with a negative book value of equity.

Bornholt (2007) introduces an approach that is called as reward beta approach to estimate expected returns and compares its performance with the CAPM and the three factor model by using the same data of Fama French (1992)’s study. The

(39)

25 reward beta approach introduced in this paper uses forward-looking portfolio reward beta estimates in the following equation to estimate expected returns.

] r ) R ( E [ r ] R [ E _i _f E_ri _m _f , for all idN

Then a version of the market model that is compatible with the reward beta approach is constructed. The version of market model is:

, )] R ( E R [ ] r ) R ( E [ r R_j _f E_r_j _m _f E_j _m _m H_j

In this model, portfolio j’ s expected return is determined by its reward beta, βrj, the risk-free rate and the market risk premium. The model’s error specification implies that βj in the equation above equals the portfolio’s CAPM beta.

This study compares the CAPM, the Fama–French three-factor model and the reward beta approach by using the same data and portfolio formation procedures with the study of Fama French (1992). Bornholt (2007) evaluates the three competing approaches by splitting the overall sample into two parts: within-sample period in which each model’s parameters’ time-series estimates are calculated and out-of-sample in which time series estimates are used in cross-section regressions to test the competing models’ explanations of portfolio out-of-sample average excess returns.

The empirical evidence in this paper does not support the CAPM and the three factor models. On the other hand, the reward beta approach is strongly supported by the empirical evidence reported in this paper.

Ho and Hung (2009) investigate whether incorporating investor sentiment as conditioning information in asset-pricing models helps capture the impacts of the size, value, liquidity and momentum effects on risk-adjusted returns of individual stocks.

(40)

26 Their findings indicate that investor sentiment plays an important role in conditional asset-pricing models for capturing the anomalies.

In Appendix-B the above- mentioned studies about multifactor models for the developed markets are given.

1.2.2.2.2 Studies for Emerging Markets

The effects of ignored firm specific factors by the CAPM on the variation of stock returns for the emerging markets have also been examined. In this part, the studies for the Istanbul Stock Exchange, the only securities exchanges in Turkey, will be addressed under the heading of Studies for Turkey.

Rouwenhorst (1999) examines the sources of return variation in emerging stock markets. They try to answer two sets of questions. The first set of three questions concerns the existence of return premiums: (1) Do the factors that explain expected return differences in developed equity markets also describe the cross section of expected returns of emerging market firms? (2) Are the return factors in emerging markets primarily local or do they have global components as well? (3) How does the emerging market evidence contribute to the international evidence from developed markets that similar return factors are present in markets around the world?

The second set of questions relates to the interpretation of the return factors. (4) Is there a cross-sectional relationship between liquidity and average returns in emerging markets? (5) Are the return factors in emerging markets cross-sectionally correlated with liquidity?

In this study, 20 emerging markets with 1750 individual stocks are analyzed during the period 1982 to 1997. As a result, it is found that the return factors in emerging markets are qualitatively similar to those in developed markets. Small stocks outperform large stocks, value stocks outperform growth stocks and emerging

(41)

27 markets stocks exhibit momentum. He reaches to the same results for Turkish market. Portfolios with small stocks have more return than portfolios with large stocks (0.72 % monthly), and portfolios of value stocks have more return than portfolios of growth stocks (2.86 % monthly) in Turkish market. There is no evidence that local market betas are associated with average returns. The low correlation between the country return factors suggests that the premiums have a strong local character. A Bayesian analysis of the return premiums in developed and emerging markets shows that, unless one has strong prior beliefs to the contrary, the empirical evidence favors the hypothesis that size, momentum, and value strategies are compensated for in expected returns around the world.

Ho, Strange and Piesse (2000) test the independent and joint roles of market, size, book to market, market leverage, book leverage, dividend yield and ea rnings to price factors on the variations of stock returns for Hong Kong stock market during the period of January 1980 to December 1994. They use an extended version of Fama and French’s (1992) cross-sectional estimation model:

i 1 t , i 8 1 t , i 7 1 t , i 6 1 t , i 5 1 t , i 4 1 t , i 3 1 t , i 2 p 1 0 u ) P ln( ) DY ln( ) P / E ln( ) BE / A ln( ) M E / A ln( ) M E / BE ln( ) M E ln( Rit J J J J J J J E J J

where Rit: excess return on stock i, βp: post-ranking beta for portfolio p, ME: market equity, BE: book equity, BE/ME: book to market equity, A: total asset, A/ME: market leverage, A/BE: book leverage, E: earnings pers hare, E/P: earnings – price ratio, DY: dividend yield, P: price pers hare of the stock, ln: natural logarithm.

They follow four steps to testing the model. In the first step, they form portfolios based on firm size and then beta for the first two years of the analysis period. In the next step, they estimate post ranking beta for these constructed portfolios for the remaining part of the analysis period. Then they adopt the Fama MacBeth (1993) algorithm and hypotheses testing.

(42)

28 Their results indicate that beta, book leverage, earnings-price ratio and dividend yield are not priced, whereas relatively strong book-to-market equity and market leverage effects and marginally significant size and share price effects are observed.

Drew and Veeraraghaven (2002) examine whether there is a size and value premium in markets outside the USA and the multifactor model of Fama-French (1996) can capture the cross-section of average stock returns for the Malaysian setting. In this study, the firms with available returns data during the period December 1992 to December 1999 are used. They construct six portfolios as the intersection of the two size portfolios and three book to market portfolios and analyze the relationship between the excess returns of these portfolios’ excess returns with market, size and value risk premium by using Fama French Three Factor model. They find that size and value premium exist in markets outside the USA and small and high book-to-market equity stocks generate higher returns than big and low book-to-market equity stocks in the Malaysian setting.

Empirical researches investigating the effects of firm specific factors on the stock returns of ISE firms have conflicting results. Most of the important studies that use directly ISE database are given below.

Aydogan and Gursoy (2000) investigate the ability of average P/E and book-to-market ratios to predict future stock market returns in emerging equity markets during the period of 1986-1999. They use a group of countries widely known as "emerging equity markets" defined and monitored by IFC arm of the World Bank. They compute the market averages of E/P and BE/ME for all analyzed emerging markets and pool them to see whether they are related with 3, 6 and 12-month future returns. They make groups based on ranking pooled observations with respect to E/P and BE/ME and evaluate their relations between various future returns. They find that returns are higher after observing a high E/P (low BE/ME) in a market. They test statistical relationship between E/P, BE/ME and future returns by using two sets of econometric tests. They initially group observed average E/P and BE/ME into